¹ Handley, David, ‘Public Opinion and European Integration’, European Journal of Political Research, IX (1981), 335–64.CrossRefGoogle Scholar

² Jowell, Roger and Hoinville, Gerald, eds, Britain into Europe (London: Croom Helm, 1977)Google Scholar; Berrington, Hugh, ‘Public Opinion and the Common Market’, paper presented at the Universities Association for Contemporary European Studies, 1975.Google Scholar

³ Jowell, and Hoinville, , Britain into EuropeGoogle Scholar; Berrington, , ‘Public Opinion and the Common Market’.Google Scholar

⁴ Deutsch, Karl and Merritt, Richard, ‘Effects of Events on National and International Images’, in Kelman, H., ed., International Behavior (New York: Holt, 1965)Google Scholar; Rosenau, James, Public Opinion and Foreign Policy (New York: Random House, 1961).Google Scholar

⁵ Butler, David and Stokes, Donald, Political Change in Britain: The Evolution of Electoral Choice, 2nd edn (London: Macmillan, 1974), pp. 309–13.CrossRefGoogle Scholar

⁶ Abravanel, M. and Hughes, Barry, ‘Public Opinion and Foreign Policy Behavior’, in McGowan, Patrick, ed., Sage International Yearbook of Foreign Policy Studies (Beverly Hills, Calif.: Sage Publications, 1973)Google Scholar; Mueller, John, War, Presidents and Public Opinion (New York: Wiley, 1973)Google Scholar; Rosenau, , Public Opinion and Foreign PolicyGoogle Scholar; Shepard, Robert, Public Opinion and European Integration (Lexington, Mass.: Lexington, 1975).Google Scholar

⁷ Peterson, Sophia, ‘Events, Mass Opinion and Elite Attitudes’, in Merritt, Richard, ed., Communication in International Politics (Champaign-Urbana, Ill.: University of Illinois Press, 1972), pp. 252–71.Google Scholar

⁸ MacKuen, Michael, ‘Social Communication and the Mass Policy Agenda’, in MacKuen, Michael and Coombs, Steve, More Than News (Beverly Hills, Calif.: Sage Publications, 1981).Google Scholar

⁹ Klapper, Joseph, The Effects of Mass Communication (New York: Free Press, 1960).Google Scholar

¹⁰ Berrington, , ‘Public Opinion and the Common Market’.Google Scholar

¹¹ These data were compiled from the Eurobarometer report and the Gallup Opinion Index. We especially would like to thank Jacques-Réné Rabier and Norman Webb for their assistance in compiling these trends.

¹² Jowell, and Hoinville, , Britain into Europe, Chap. 2.Google Scholar

¹³ Jowell, and Hoinville, , Britain into Europe, pp. 54–7.Google Scholar

¹⁴ Lodge, Juliet and Herman, Valentine, ‘Direct Elections to the European Parliament’, European Journal of Political Research, VIII (1980), 45–62.CrossRefGoogle Scholar

¹⁵ Eurobarometer Reports (Brussels: Commission of the European Communities, semi-annual series).

¹⁶ The Times might have been a preferable source, and one which has been used by other events-data projects. However, The Times was closed by a strike for a crucial six-month period in 1979.

¹⁷ Seymour-Ure, Colin, ‘Press’, in Butler, David and Kitzinger, Uwe, eds, The 1975 Referendum (London: Macmillan, 1976).Google Scholar

¹⁸ Rose, Richard, Politics in England, 3rd edn (Boston: Little, Brown, 1980), pp. 214–16.Google Scholar

¹⁹ Each Tuesday, Wednesday and Thursday edition of The Guardian was examined for all articles explicitly dealing with Britain's relationship with the European Communities or the impact of European integration on British interests. A total of 3,788 articles appeared during this time period. The events data presented in this research are based only upon news articles and political advertisements during the 1975 referendum (N = 3,045); this excludes editorials and feature stories.

Each article was treated as the unit of analysis. Based on past studies of events data, we developed a scheme for coding the visibility of an article, its evaluative content, the actors involved and similar factors. When more than one event or actor was involved in an article, the coder determined the primary actor for the actor-related variables. However, overall evaluative content was always based on the total content of the article. Detailed information on the coding procedures is available from the authors. Initial coding of all variables was verified by the study directors.

²⁰ Because of the Sun's limited coverage of international issues, this correlation has been corrected to adjust for the low reliability of Sun scores in months with very few sampled articles. Time-points were weighted by the volume of the Sun's coverage (total column inches) during the month. A simple unweighted comparison still produces a 0·52 correlation between both media series.

²¹ The balance of Community-related news is computed by multiplying the score for the overall evaluative content of the article by the number of column inches for the article. The resulting statistics were summed to produce monthly totals for the balance of news stories.

²² Särlvik, Bo et al. , ‘Britain's Membership in the EEC’, European Journal of Political Research, IV (1976), 83–113Google Scholar; Handley, , ‘Public Opinion and European Integration’Google Scholar; Inglehart, Ronald, The Silent Revolution (Princeton: Princeton University Press, 1977), Chap. 12.Google Scholar

²³ Abravanel, and Hughes, , ‘Public Opinion and Foreign Policy’Google Scholar; for another application of this logic see Converse, Philip, ‘The Concept of the Normal Vote’, in Campbell, Angus et al. , Elections and the Political Order (New York: Wiley, 1966).Google Scholar

²⁴ Abravanel, and Hughes, , ‘Public Opinion and Foreign Policy’Google Scholar; Mahoney, R. B., ‘The Superpower Balance, Military Policy and Public Opinion in the United Kingdom, France, and the Federal Republic of Germany’, in Daniel, D., ed., International Perceptions of the Superpower Balance (New York: Praeger, 1978)Google Scholar; Peterson, , ‘Events, Mass Opinion, and Elite Attitudes’.Google Scholar

²⁵ MacKuen, , ‘Social Communication and the Mass Policy Agenda’Google Scholar; Erbring, Lutz, ‘The Impact of Political Events on Mass Publics’ (Ann Arbor: doctoral dissertation, University of Michigan, 1975).Google Scholar

²⁶ A model of two-step opinion flow also was tested with these data using the procedures described in MacKuen, , ‘Social Communication and the Mass Policy Agenda’, pp. 37–46Google Scholar. The predictive power of the two-step model equalled the opinion-decay model, even though the former includes more parameters. Parsimony led us to rely on the simpler, though equally accurate, opinion-decay model as the best representation of the events-opinion linkage.

²⁷ The ‘good thing’ question was used to establish the public opinion trend when this item was available. Otherwise, the linear relationship between PDI scores for the two questions was estimated by comparing the results when both questions were used in the same-survey. If only the ‘feel sorry’ question had been available for a month, this linear relationship was used to transform the ‘feel sorry’ PDI score to the ‘good opinion’ metric. When both questions were available their PDI scores were averaged. The same procedure was used to add the time-points from Table 1 to the opinion time series. The total N for the opinion series is 42. The events series is based on the time series of Community-related news stories presented in Figure 2.

²⁸ The dynamic model utilized in this study had been developed by MacKuen, (‘Social Communication and the Mass Policy Agenda’)Google Scholar and Erbring, (The Impact of Political Events)Google Scholar. The basic elements of their model can be described briefly, although the interested reader is encouraged to consult the original sources.

Two major problems arise in the estimation of the model's parameters. Firstly, the data for the opinion time series are unequally spaced. Secondly, it is quite likely that the error terms in the model are serially correlated. Fortunately a single solution is available for both of these problems. A two-stage iterative process is employed to supply estimates of lagged Y values. Since predicted values do not contain their residual errors, the problem of serially correlated errors has been purged a priori through this use of predicted Y values. In addition, these predicted Y values are generated for all time-points, so that first-order differences can be calculated for every month.

The iterative estimation process is conceptually straightforward. The first stage of the process begins with an initial estimate of the model's parameters (b and c), and uses the first values of X and Y at time t − 1 to predict Y _t. Using the predicted Y value and the next X value, the lagged Y values are sequentially estimated for the entire series. In the second stage these estimates of the lagged Y values are employed with the actual y values to obtain new estimates of the model. A Marquardt non-linear least squares procedure is used to estimate the parameters b and c in order to improve the overall fit of the model. These new parameters are used to repeat the first-stage iterative process and estimate new lagged Y values. This two-stage process is continued until the RSS for the model cannot be reduced further.

²⁹ The reciprocal of c (3·32) represents the time-lag in months required for 63·2 per cent of the initial and residual impact of an event to be accumulated (see MacKuen, , ‘Social Communication and the Mass Policy Agenda’, p. 45Google Scholar). Thus, opinions do lag behind events, but the time-lag is relatively brief (cf., Abravanel, and Hughes, , ‘Public Opinion and Foreign Policy Behavior’, p. 121).Google Scholar

³⁰ MacKuen, , ‘Social Communication and the Mass Policy Agenda’, Chap 5.Google Scholar

³¹ Butler, and Stokes, , Political Change in Britain, p. 279Google Scholar; Jowell, and Hoinville, , Britain into Europe, Chap. 3.Google Scholar

³² The results of these analyses were: parameter b, coefficient 5·47, standard error 1·57; parameter c, coefficient 0·37, standard error 0·28; N = 23, R ² = 0·65.

³³ We calculated the average volatility (standard deviation) of PDI scores for the ‘good thing’ question from 1973 until 1981 for all EC members except Luxembourg. The volatility of responses in Britain (14·93), Ireland (9·41) and Denmark (7·41) exceeds the average for the original members (6·99).

³⁴ Miller, Arthur H., Goldenberg, Edie N. and Erbring, Lutz, ‘Type-set Politics’, American Political Science Review, LXXIII (1979), 67–84.CrossRefGoogle Scholar

³⁵ Berrington, , ‘Public Opinion and the Common Market’.Google Scholar

³⁶ Mueller, , War, Presidents and Public Opinion.Google Scholar

³⁷ Inglehart, Ronald, ‘An End to European Integration?’ American Political Science Review, LXI (1967), 91–105CrossRefGoogle Scholar; Inglehart, , The Silent RevolutionGoogle Scholar; Lindberg, Leon and Scheingold, S., Europe's Would-be Polity (Englewood Cliffs, N.J.: Prentice Hall, 1967)Google Scholar; Shepard, , Public Opinion and European Integration.Google Scholar

³⁸ Handley, , ‘Public Opinion and European Integration’.Google Scholar